interpersonal relationships and child development

DEVELOPMENTAL REVIEW 7, t-21 (1987)

Interpersonal Relationships and Child Development

ROBERT A. HINDEANDJOAN STEVENSON-HINDE

MRC Unit on the Development and Integration of Behaviour, Madingley, Cambridge, United Kingdom

In studies of psychological development, the child must be seen not as an isolated unit, but as a social being, forming part of a network of relationships. Inter- actions, relationships, social groups, and the sociocultural structure form successive levels of social complexity, each level involving properties not relevant to lower levels. The levels are connected by dialectical relations and are to be seen not as entities but as processes in continuous creation through the agency of the dialectics. It is rarely possible to study one level in isolation; the dialectics almost always obtrude. A relationships approach must be integrated with others in the held, and especially with that of family systems theorists. G 1987 Academic Press. Inc.

The nature and dynamics of interpersonal relationships impinge on many branches of psychology, but until relatively recently they have not been the focus of systematic study. During the last decade the necessity for a concerted effort in this area has been recognized by a number of authors (e.g., Duck & Gilmour, 1981; Hinde, 1979; Kelley, 1979; Kelley et al., 1983). However, the implications of a relationships approach are not yet widely recognized. The purpose of this article is to spell out some of those implications, especially as they apply to the field of child development .

In our society, most children grow up in a nuclear family, interacting daily with one or two parents, with siblings, and from time to time with relatives and friends. In due course, through schools and peer groups, the range and variety of their interactions increase (e.g., Foot, Chapman, & Smith, 1980; Rubin & Ross, 1982). In other societies the details differ (e.g., Whiting & Whiting, 197.5), but the general pattern of a network of relationships expanding from one or more with blood relatives (including nearly always the mother) is ubiquitous. Indeed, something similar occurs in many nonhuman primates (Berman, 1983) and was almost certainly present in, to use Bowlby’s (1969) term, our “environment of evo- lutionary adaptedness” (Alexander, 1974; Mellen, 1981; Short, 1979).

This network of interpersonal relationships, of which the growing child forms part, constitutes a crucially important part of his or her environment. In studying psychological development, therefore, it is necessary to treat the child not as an isolated entity but as a social being, formed by and forming part of a network of relationships which are crucial to its

This work was supported by the Royal Society and the Medical Research Council.

0273-2297187 $3.00 Copyright 0 1987 by Academic Press, Inc. All rights of reproduction in any form reserved.

2 HINDE AND STEVENSON-HINDE

integrity (see, e.g., Bronfenbrenner, 1979; Clarke-Stewart, 1978; Parke, Power, & Gottman, 1979). This view is now becoming widely accepted, and Child Devefopmenf (1985, 56(2)) recently devoted nearly a whole issue to studies of children in a family context. Thus it seems timely to focus on the special problems posed by relationships to the developmental psychologist.

SOME GENERAL ISSUES-INTERACTIONS, RELATIONSHIPS, AND GROUPS

It is necessary first to consider the nature of relationships in general terms. Most data on social behavior concern interactions between individuals. The developmental psychologist, though focusing on a particular child, is recording behavior generated by two or more individuals- peers, parent and child, teacher and child, and so on. Even in a test situation, the behavior observed is in part a product of interaction with the experimenter (Perret-Clermont & Brossard, 1985).

When two individuals interact on successive occasions over time, each interaction may affect subsequent ones, and we speak of them as having a relationship. Their relationship includes not only what they do together, but the perceptions, fears, expectations, and so on that each has about the other and about the future course of the relationship, based in part on the individual histories of the two interactants and the past history of their relationship with each other. For instance, a mother’s behavior may depend not just on her child’s behavior at the moment, but also on their past interactions (Halverson dc Waldrop, 1970) and on her hopes for the future.

The nature of any interaction depends on the characteristics of both (or all) of the individuals involved. For example, Elder, Van Nguyen, and Caspi (1985) found that drastic income loss was associated with rejecting behavior of fathers (but not mothers) to daughters (but not sons). How- ever, the extent of the effect was influenced by a characteristic of the other partner in the relationship, namely the unattractiveness of the daughter. Since interactions and relationships depend on both participants, data obtained from observation of interactions cannot be ascribed solely to the characteristics of one or the other participant. Thus how quickly a mother goes to a crying baby is not solely a measure of her sensitivity, but depends in part on how often the baby has cried recently. And how often a baby cries depends in part on how quickly the mother goes to it when it does so.

While the nature of an interaction depends in part on the individuals involved, it is affected also by the relationship in which it is embedded. Thus the behavior each individual shows depends in part on his or her feelings and expectations about the relationship of which the interaction forms part. The extent to which each placates the other, demands, dis-

INTERPERSONAL RELATIONSHIPS 3

plays affection, and so on, will be partially influenced by his hopes for the future. In the longer run the behavior an individual can show depends in part on the relationships he or she has experienced in the past. While early relationships may have a crucial importance, it is probable that current relationships continue to affect future responsiveness throughout life. And the nature of the relationship, and the participants feelings and predictions about it, depend on the nature of the interactions. Thus we must come to terms with two dialectics-between the characteristics of individuals and interactions on the one hand, and between interactions and relationships on the other, with two-way cause-effect influences in each case.

However, that is not all (see Fig. 1). Each relationship is influenced by the social nexus of other relationships in which it is embedded: A’s relationship with B is affected by B’s relationship with C, and so on. Con- versely the characteristics of the social group are determined by the dyadic and higher order relationships within it. Relationships are affected also by the social norms and values current in the group: the dynamics of a marriage, or of a parent-child relationship, are affected by the expectations and goals of the participants and by the relations between those expectations and perceived reality (e.g., Andreyeva & Gozman, 1981). Those norms and values are transmitted and transmuted through the agency of dyadic relationships.

And the sociocultural structure, used here for the system of institutions and beliefs, and the relations between them, shared by the members of the group, in turn both influences and is influenced by the relationships between individuals (e.g., Hinde, 1984). To give but one example, individual attitudes toward males vs females is related to the presence of a female deity in the belief system of the group (Williams & Best, 1982). Beyond that, each group is juxtaposed with other groups, contact with which affects diverse aspects of the social behavior of its individuals. Finally each group is set in a physical environment, which affects and is affected by the group members. Social scientists must therefore come to terms with a series of dialectics between successive levels of complexity -the behavior of individuals, interactions, relationships, social struc-

FIG. 1. The dialectics between successive levels of social complexity (modified Hinde, 1979).

from


ture, sociocultural structure, and intergroup relationships. At the same time they must remember that each level represents not an entity but a process in continuous creation through the agency of the dialectics.

Each of these levels has properties not relevant to the level below. For example, synchrony and meshing are relevant to interactions but not to the behavior of individuals; properties of relationships that depend on the relative frequency or patterning of different types of interactions (e.g., is maternal control associated with nurturance and warmth?) are not relevant to individual interactions; and a group may have characteristics of internal structure (hierarchical, centripetal, etc.) irrelevant to the component relationships. Furthermore, as we emphasize later, properties at more than one level may be important for understanding dynamics: for instance, the course of interactions may be affected by the participant’s perceptions of the relationship.

While some of the social sciences seem to be concerned with one or another of these levels, in practice the dialectics always obtrude. It is rarely possible to study one level in isolation. Thus at the individual level, students of personality find that the cross-situational consistency of supposed “traits” tends to be low. While some of the weakness in the correlations is due to measurement error, and can be overcome by aggregation (Block, 1981; Rushton, Brainerd, & Pressley, 1983), it is necessary to recognize that behavior may be affected (to differing extents according to the nature of the individual and of the behavior) by the context (Bern & Funder, 1978; Endler & Hunt, 1968, 1969; Endler & Magnusson, 1976; Kenrick & Stringfield, 1980; Mischel, 1973). And the most important aspect of the context is the interactional and relationship one, including the meaning that the individual attributes to his relationships according to his or her sociocultural scheme of reference and past experience.

Developmental psychologists are now well aware that, for instance, 12- to 18-month-olds may behave differently when with their mothers in the Ainsworth Strange Situation from when with their fathers (Grossmann, Grossmann, Huber, & Wartner, 1981; Main & Weston, 1981; see also Sroufe, 1985), and that 4-year-olds adjust the language they use according to whom they are with (Gelman & Shatz, 1977; Shatz & Gelman, 1973; Snow, 1972). They have had to come to terms with the interacting influences of parent on child and child on parent (e.g., Bell & Harper, 1977), and to consider the relative importance of parent-child and peer relationships in the development of personality (e.g., Bowlby, 1969; Sullivan, 1938; Youniss, 1980).

Among social psychologists, the same issue is implicit in the symbolic interactionists’ description of individuals as having a number of “role identities” which emerge in interaction with a particular other and provide plans for action and criteria for evaluating action (Goffman, 1959; G. J. McCall, 1974; M. McCall, 1970). The interdependence theorist


makes a related point in emphasizing that every relationship is in part a product of its own history and anticipated future and is thereby differen- tiated from every other relationship (Kelley, 1979).

Even cognitive psychologists find that the mathematics used in the market place may differ from those used in the schoolroom (Carraher, Carraher, & Schliemann, 1985), and that how an individual tackles an intellectual problem may change radically with the social situation (Doise & Mugny, 1984; Donaldson, 1978; Perret-Clermont & Brossard, 1985). And anthropologists and sociologists, concerned with sociocultural structure, seek to understand the ways in which beliefs, myths and legends affect the lives of individuals and reciprocally how those beliefs reflect the natures, desires, wishes, and frustrations of individuals (e.g., Herdt, 1981; Keesing, 1982).

THE TWO ROUTES FOR GENERALIZATIONS FROM INTERACTIONS

As we have seen, data about social behavior concern interactions. From data on interactions there are two principle routes to generalizations (Hinde & Stevenson-Hinde, 1976). These can be illustrated by reference to Fig. 2, which concerns data on two-parent, one-child families.

of

INTERACTIONS

3

1 ,b-,

FAMILIES

RELATIONSHIPS

FIG. 2. The relations between interactions, relationships, and families, illustrating the two routes for generalizations from data on interactions (modified from Hinde & Ste- venson-Hinde, 1976).


In that figure, Rectangles 1, 2, and 3 represent specific instances of three types of interaction between a particular mother and her infant-say mother inhibits and child complies, mother inhibits and child ignores, and mother speaks affectionately. Such are the basic data. From them we can make generalizations about the incidence of these three types of interaction in that mother-infant dyad (4, 5, and 6). The rectangles behind 4 represent comparable data on maternal inhibitions with child compliance for other mother-infant dyads, those behind 5 for maternal inhibition without compliance and those behind 6 for maternal affection. The usuaI route is to remain at the same level of social complexity, seeking for generalizations about each type of interaction across dyads, as at 7, 8, and 9. This is the procedure followed in most studies of the relations between particular independent variables (e.g., sex or age of child) and particular dependent variables (e.g., maternal affection). Comparable procedures could be followed for interactions between father and infant, between mother and father, and for triadic interactions.

A second route involves proceeding to the relationship level of social complexity, examining the incidence of and relations between different types of interaction in the same dyad. Thus in the figure, 10 represents aspects of this particular mother-infant relationship (and the rectangles behind it comparable aspects of other mother-infant relationships). These will include not only the frequencies of the different types of interactions, but also their relative frequencies (e.g., were expressions of affection more or less common than maternal inhibitions), the sequential relations between different types of interaction (e.g., were maternal inhibitions accompanied by expressions of affection, or were these two maternal modes always separated in time), and other properties (see below).

Comparable procedures allow generalizations about other relationships in that family, i.e., father-infant (11), mother-father (12), and their triadic relationship (13), and in other families.

From here we can again remain at the same level, seeking generalizations about mother-infant relationships for a number of dyads (14), father-infant relationships (15), etc.; or we can proceed to the group level, in this case examining the nature and structure of that particular family (16). The latter would include the relations between the intrafamilial relationships. This could lead in turn either to generalizations about families (17) or to descriptions of higher order groups within which the families were located (not shown).

It must be noted that although Fig. 2 portrays the task of description as proceeding from interactions to relationships and family structure, it can sometimes proceed in the opposite direction: that is, descriptions of relationships can permit deductions about the nature of interactions. More importantly, cause-effect relations proceed in both directions (see above and Fig. 1).


The diagram emphasizes that attempts to obtain generalizations about relationships (14, 15) from generalizations about specific interactions (7, 8, & 9) would lose information about some of the most important properties of relationships, namely those that depend upon the relative frequency and patterning of different types of interactions. The number and the proportions of maternal inhibitions complied with provide information about very different aspects of the mother-child relationship, and a given number of commands from a mother who also often expresses affection is unlikely to have the same meaning for the child as the same number of commands from a mother who never expresses affection would have. Such properties of the mother-child relationship would be lost if we examined merely the absolute frequencies of different types of interactions.

Similarly, attempts to obtain generalizations about families (17) from generalizations about relationships (14, 15) would lose crucial information about how relationships are patterned and affect each other within families. A mother-child relationship involving frequent inhibitions might have quite different impacts on the child, depending on whether or not the father-child relationship was also inhibitory. Furthermore the properties of the mother-child relationship may affect the father-child relationship, and the manner in which it does so may depend on the nature of the parental relationship. Such issues would be lost if we merely examined the properties of the several relationships.

Of course, which of the approaches represented in Fig. 2 is more appropriate depends upon the problem. For example, studies of the effects of given independent variables upon particular types of behavior often involve generalizations across dyads (7-9). Where dyadic differences are important or are the focus of study, and one type of interaction is salient, it may be adequate to use differences at the interaction level (e.g., 4, 5, or 6) as an index of the relationship. But interaction measures (e.g., frequency of maternal controls) may be less revealing than the relations between interactions (e.g., proportion of maternal controls that are successful, proportion of interactions that are positive). And where more than one aspect of the relationship, or the relations between different types of interaction within the relationship, are important, the comparisons are at the relationship level (e.g., IO). For instance, the distinctions between authoritarian, authoritative, and permissive parents depend upon assessments of several types of interaction concerned with parental control and acceptance (Baumrind, 1967, 1971). This type of approach (reviewed in Maccoby & Martin, 1983) has been more successful than those involving simple aspects of parent-child interaction.

This suggests that it is sometimes more accurate to think of personality as influenced by relationships than by specific types of interaction. This could be critically important in many ways. For instance, if a single


mother attempts to fill two roles by indulging in rough-and-tumble play with her son as well as by behaving with maternal solicitude and sensitivity, are the consequences of rough-and-tumble play coming from the same person as she who provides maternal sensitivity the same as if it were part of a different relationship? We do not know the answer to this, but thinking in relationship terms at least forces us to put the question.

But we must recognize that relationships can never be adequately assessed along a single scale. Of course, the same is true of interactions: aggression must be assessed not only in terms of frequency, but also of quality, context dependence, and so on. But the problem is very much more severe with relationships. Relationships are measurable along many dimensions, and some of the most important may involve ratio measures (Hinde & Herrmann, 1977). The proportion of occasions on which a baby cries in which the mother picks it up tells us something different about their relationship from the absolute frequencies of crying or picking up. At best we can categorize relationships according to where they stand along a limited number of dimensions that we deem to be important.

THE DESCRIPTION OF RELATIONSHIPS

Although difficult, the task of describing relevant features of relationships must not be bypassed. The generalizations we make, or the explanatory concepts we use, will depend upon the sort of relationship with which we are concerned.

Major decisions implicit in any system for describing relationships concern the level of analysis at which to proceed and the sort of categories likely to be useful. One approach is to select initially the level of analysis at which we habitually talk about relationships, and aspects of relationships that appear to be important in everyday life. Such an approach can be justified by the view that we have been shaped, culturally and/or biologically, to be reasonably efficient prognosticians about relationships (e.g., Jolly, 1966; Humphrey, 1976; Osgood, 1969). Thus, as a way of ordering the almost limitless data about relationships, it has been suggested (Hinde, 1979) that the more important dimensions fall into eight categories. These categories proceed from those concerned primarily with individual interactions to those concerned with more global properties of the relationship, and from its more behavioral to its more subjective aspects.

i. The content of the interactions (what do the participants do together?);

ii. The diversity of the interactions (how many different things do they do together?);

iii. The qualities of the several types of interaction;


iv. Qualities that emerge from the relative frequency and patterning of the interactions (e.g., not how often does the child do what it is told, but on what proportion of the occasions on which it receives an instruction does it comply);

v. The reciprocity vs complementarity of the various types of interactions (i.e., do the partners behave similarly to each other, or do they behave in differing but complementary fashions? This includes the question of relative power);

vi. Intimacy (how much do they reveal themselves to each other?); vii. Interpersonal perception (does A perceive B as B is, as B per-

ceives B, as close to A’s ideal partner, etc.); viii. Commitment to continuing the relationship and/or to maintaining

or improving its quality.

Not all these categories of dimensions will be relevant to any one problem: complete descriptions of a relationship is impracticable and un- necessary. But they may provide convenient pigeonholes for ordering data on relationships.

However, as emphasized already, a relationship is not a static entity but a process in continuous creation through time. Thus any description must refer to a limited span of time, and we must not forget that the future course of a relationship may be affected by events before the pe- riod in which it was studied. Indeed changes in dimensions within these categories may be as important for prognosis as the dimensions themselves.

The categories listed above are concerned with dimensions that are seen as important in everyday life. Although in large part based on data referring to a finer level of analysis (e.g., interactions, perceptions), they involve also higher level properties emergent from their patterning (Hinde, 1979). Another approach, that of Kelley et al. (1983), rests more exclusively on the interaction level of analysis. Picturing a relationship as the interconnections between the temporal chains of two individuals’ affect, thought, and action, they suggest analysis of relationships in terms of(i) the kinds of events in each chain that are interconnected and (ii) the pattern; (iii) strength, (iv) frequency, and (v) diversity of interconnections; (vi) the extent to which the interconnections facilitate or interfere with the chains of action etc; (vii) their symmetry vs asymmetry (i.e., whether the effects of each chain on the other are similar or different); and (viii) the duration of interactions and/or the relationship.

These two approaches to description are not incompatible: that of Kelley et al. (1983), attempts to specify the events upon which some of the more global properties, described in the first approach, depend. Dis- cussion at the interaction level may often be necessary to explain the


dynamic importance of some of the more global properties of relationships. However, the interaction level cannot be sufficient by itself because, apart from anything else, we evaluate relationships in terms of global properties, and the evaluations by the participants affect their relationship’s future course.

TEASING APART THE ROLES OF THE TWO PARTNERS

We have seen that the nature of an interaction depends on both (or all) participants. We must also ask how far it is meaningful, and how far it is possible, to tease apart their roles. Some discussions of this issue by developmental psychologists fall into the trap of causal chain thinking: one must never forget that the links in a supposed causal chain are in reality spiders’ webs (Hanson, 1955).

Just because at least two individuals participate in every interaction or relationship, and each is continuously influencing the other, simple answers to questions of the type, “Who is influencing whom?” are seldom meaningful. Many of the flaws in attempts to answer such questions have been reviewed by Maccoby and Martin (1983). In addition, behavior within a relationship usually involves long sequences of interactions, so that specifying the initiation may be arbitrary.

However, we can ask hard questions about differences or changes: the crucial issue is that the questions must be specified precisely. Given groups of dyads, one approach involves the use of cross-lagged analyses (e.g., Clarke-Stewart, 1973). This can perhaps tell us something about the relative extent to which interactions type A and interactions type B at Time 1 affect interactions at Time 2, but its shortcomings are now well- known (Eron, Huesmann, Lefkowitz, & Walder, 1972; Kenny, 1975; Ro- gosa, 1980). In any case it must be emphasized that the conclusions usually properly refer to influences of properties of the interactions or relationship at Time 1 and not, as is often implied, to those of the individual participants.

As another approach, given certain assumptions about how one type of interaction affects another within the relationship, some progress can be made in teasing apart the roles of the partners if we examine changes or differences in the relative frequencies of interactions of different types. For example, we might be interested in whether a change in time spent together by a mother and toddler was due primarily to a change in the mother or to a change in the toddler. If the time together increased, but a smaller proportion of approaches between mother and toddler (and a higher proportion of leavings) were due to the mother, and if the importance of long-distance signaling, etc., can be disregarded, we could de- duce that the change was due more to a change in the toddler than to a change in the mother (see Table 1). In this way we can answer questions


TABLE 1 ASSESSING THE EXTENT TO WHICH CHANGES IN TIME SPENT TOGETHER BY A MOTHER

AND TODDLER ARE DUE TO CHANGES IN THE MOTHER OR CHANGES IN THE CHILD

Behavioral measures

Types of change

Mother-child -+ Mother-child

Proportion of approaches Time together by mother

+ +

- c

Mother-child c

Mother-child +

+

- +

Nofe. The four rows concern four possible types of change (mother becomes more possessive; mother becomes less possessive; child becomes more dependent; child becomes less dependent). The symbols (+ , -) indicate the direction of change in these two measures. If the change is due to a change in the mother, these two measures should change in the same direction; if due to the child, the measures should change in opposite directions. A precisely similar argument can be applied to differences between mother-toddler dyads: If the difference is primarily due to a difference between mothers, the dyad spending more time together would have a higher proportion of approaches by the mother. If the difference is due primarily to a difference between the toddlers, the reverse would be the case.

of the type, “Is a given change in A,‘s relationship with B, due more to a change in A, or to change in B,,” or, “Is the difference between Ai’s relationship with B, and A,‘s relationship with B, due more to differences between A, and A, or between B, and B,?” The answers in each case are, it will be noted, in relative terms. Furthermore, this approach dem- onstrates the crucial importance of distinguishing questions about re- sponsibility for the nature of the relationship over a limited span of time, from those concerning changes in the relationship over time, and each from those concerning differences between relationships at one time. The method has been discussed in more detail elsewhere (Hinde, 1969, 1979), and can be extended to the triadic situation (Hinde, 1977), but has so far been applied only to nonhuman species.

Concerned not with particular cases but with generalizations, a more direct route for the developmental psychologist is to study the extent to which differences between relationships are associated with differences in the characteristics of the participating individuals. This is, of course, the route followed in studies of sex differences-for instance, in comparisons between mother-child and father-child relationships (Lamb, 1976), or between the friendships of boys and those of girls (Hartup, 1983). However, most such studies involve comparisons between mea-


sures of particular types of interaction-for instance, comparisons of father-child with mother-child relationships show that the former involve more physical play and less tender care, and the play of boy-boy dyads differs in many ways from that of girl-girl dyads (e.g., Clarke-Stewart, 1978; Hartup, 1983; Parke, 1979).

However, the sex of one or another participant may be associated not merely with differences in frequencies between particular types of interactions, but also with the relations between interactions, or what we may call the structure of the relationship. Thus in a recent study of pre- schoolers, most correlations between items of mother-child interaction were broadly similar for boys and girls. However, the item mother strong controls was associated with other negative items (e.g., mother and child hostile, mother and child noncomply with requests, mother inhibits) in boys but not in girls. This is in harmony with the view that strong controls were used by mothers of boys mainly in relationships that were generally tensionfilled, whereas with mothers of girls there was no such re- striction (and even some opposite tendency) (Hinde & Stevenson-Hinde, in preparation).

If the sex of one or both participants has a dramatic effect on the nature of a relationship, other characteristics are likely also to do so. Thus the developmental psychologist can compare mother-child relationships between depressed and nondepressed mothers (Pound, 1982), or assess the associations between dimensions of child temperament and mother- child interactions (e.g., Lee & Bates, 1985). In addition one characteristic of one or the other partner may affect the influence of another characteristic on an outcome measure. For instance, in a study of preschool children we found virtually no sex differences in the temperamental characteristics of the children in the sample and very few sex differences between measures of mother-child interaction. Nevertheless there were some differences in the correlations between temperamental characteristics and measures of mother-child interaction. These involved particu- larly the characteristic shy, assessed by maternal interview questions about initial withdrawal from strangers and not settling into strange situations. Interview data indicated that, whereas shy boys tended to have more behavior problems and worse family interactions than nonshy boys, for shy girls the opposite was the case. A similar picture was provided from observational data on mother-child interaction, obtained on different days by a different observer. Thus the characteristic “shy,” as measured by the instruments we used, interacts with sex to affect the mother-child relationship. Comments made by a number of mothers in the interviews suggested that this difference was a consequence of social norms-for these mothers it was appropriate for a little girl to be shy, but not for a little boy (Simpson & Stevenson-Hinde, 1985). Thus maternal


values appear to affect many aspects of the mother-child relationship, and perhaps they have ramifying effects throughout the family (Fig. 1).

CROSSING THE LEVELS OF SOCIAL COMPLEXITY

Awareness of the dialectics between successive levels of social complexity (Fig. 1) brings recognition of the need to cross levels in order to understand the dynamics of changes or differences in relationships, or to understand mutual influences between relationships. The differences between the correlations between “shy” ratings and measures of mother- child interactions between boys and girls is a case in point: the relationship differences appear to be due to differences in social norms for boys vs girls. A more interesting example is provided in the study by Sroufe, Jacobvitz, Mangelsdorf, De Angelo, and Ward (1985) of mothers showing a “seductive” pattern with their 2-year-old sons. Such mothers tended to be not seductive but hostile (deriding) toward their daughters. Thus the “seductive” pattern cannot be viewed in maternal trait terms, but depends critically on the relationship. However, this mother-son/mother- daughter difference was understandable in terms of psychological processes in the mothers related to a history of emotional exploitation by their own fathers, and the resulting dissolution of “generational bound- aries.” And these psychological processes in the mothers appeared to involve reconstructing relationship patterns that they had known in their own families.

Similarly, Belsky and Isabella (1985) found that recollected experiences as a child in the family of origin predicted changes in marital quality following the birth and rearing of a baby: they suggest that a warm upbringing enhances communicative competence which in turn facilitates the maintenance of the marital relationship in spite of the changes in rou- tine that inevitably accompany the birth of a baby.

THE NATURE OF EXPLANATORY CONCEPTS

The fact that the behavior shown in an interaction depends on both participants forces us to consider carefully the nature of the explanatory concepts that we use. For instance, the categories of attachment derived from the Ainsworth Strange Situation (Ainsworth, Blehar, Waters, & Wall, 1978) reflect aspects of a relationship more than individual characteristics: children may be categorized differently according to the parent who is present (Grossmann et al., 1981; Main & Weston, 1981). But what about supposed measures of child characteristics, such as temperament? In so far as temperament assessments are based on the child’s behavior in particular relationships or social situations, they may reflect aspects of relationships as well as individuals. While the moderate nature of the correlations usually obtained between assessments of temperament by


fathers and mothers (reviewed in Bates, in press) could be ascribed to error, they may also reflect differences between the father-child and mother-child relationships. These might include differences in what father and mother do with the child, the quality of their interactions, the extent to which their relationships with the child are nurturant/succorant, authoritarian/permissive, etc., their differing intimacy with the child, the differences in their perceptions of the child, and in the child’s perceptions of them, in each case as assessed against the expectations and values of the perceiver, and so on.

Indeed it may be useful to regard measures of supposed child characteristics as arranged along a continuum from pure individual characteristics at one end to relationship characteristics at the other. Height and weight, but few psychological characteristics, would lie right at the individual end, attachment categories nearer the relationship end. The several temperament characteristics would lie near, but at different distances from, the individual end (Stevenson-Hinde, in press). This, of course, does not mean that relationship characteristics such as attachment may not affect characteristics lying near the individual end in the longer term.

THE DYNAMICS OF CONSISTENCY/lNCONSlSTENCY

If measures of characteristics such as temperament differ with the social situation, two courses are open to the investigator. One possibility would be to aggregate measures across situations, hoping thereby to assess a more basic mediating variable that would predict intersubject differences. The principle of aggregation depends on the view that the sum of a set of multiple measurements is a more stable and representative estimator than any simple measurement. Rushton et al. (1983) have reviewed 12 areas of developmental research in which aggregation was useful, reaching conclusions different from those suggested by the low correlations produced when individual measures were used. Such a procedure is more appropriate, the more the measure reflects an individual characteristic rather than a relationship one. However, even then such a variable might well not predict behavior in particular situations accu- rately. For instance, we might attempt to assess the aggressiveness of a number of boys by pairing them with a number of partners drawn at random from a pool. The more partners we use, the more prognostic value will a ranking of the subjects on their mean scores have. However, since the aggression shown depends on absolute and relative characteristics of both subject and partner, the mean scores may well not accu- rately predict aggression toward a new partner, and certainly will tell us little about the bases of the variance of an individual’s behavior with different partners.


A second possible course is to attempt to understand the processes underlying cross-situational differences in behavior. As indicated above, a first step here could come from an examination of the effect of one characteristic on the relations between the interactions or relationships in the two situations. For example, a comparison between behavior in preschool and the mother-child relationship showed that certain forms of aggressiveness and assertiveness in school were associated at home with maternal control and inhibitions in girls but not in boys, suggesting that mothers tolerate certain types of behavior in boys but not in girls (Hinde & Tamplin, 1983).

As another example, preschool children assessed as shy on the basis of maternal descriptions of behavior at home tended not to interact with peers in their second term in preschool. However, 8 months later some of the relations between shy ratings and behavior at preschool had altered in ways that suggested that the bases of “shyness” differed between boys and girls. There was an overall tendency for shy children not to initiate interactions with teachers, but while the interactions of shy boys with teachers contained a low proportion of neutral speech and friendly responses, those of shy girls contained a low proportion of hostile ones. Furthermore, while shy boys tended (nonsignificantly) to play interac- tively with peers, the opposite was the case with girls, who were often passive. But when shy girls did interact with peers, they (unlike shy boys) seldom displayed hostility and were seldom ignored. Thus while the shy boys tended to have a small proportion of positive interactions, shy girls tended to be passive and have few negative ones (Hinde, Ste- venson-Hinde, & Tamplin, 1985). Such findings with relatively small samples (IV = 37) must be treated with caution. However, they illustrate that, while a limited amount of aggregation is often necessary, one way forward lies not in aggregating measures to assess global characteristics, but rather in looking in more detail at more measures to understand the dynamics of what is going on. Such considerations also provide a warning against oversimplifying the continuity/discontinuity controversy. Conti- nuities may be revealed in ways other than by comparing similar patterns of behavior over time. What we must search for has been termed by Sroufe (1979) “coherence across transformations” (see also Hinde & Bateson, 1984). At one level, this involves recognizing that the same behavioral propensity may be revealed in different ways at different ages or in different situations. At another, it means that we must come to terms with the ways in which one type of experience or relationship may affect behavior of a different sort in a different context and/or at a different stage in the life history. Sroufe’s work on the precursors of seductive behavior in mothers exemplifies this (see above). As another example, in a study of the effects of institutional rearing, Rutter, Quinton, and Liddle

I6 HINDE AND STEVENSON-HINDE

(1983) found that the strongest effect on the women’s parental styles was provided by the characteristics of the women’s spouses. This, however, could be seen as an indirect effect of the institutional rearing: over half of the institution-reared women married men with psychosocial problems, as compared with 13% in the general population comparison group. Thus the childhood experiences influenced the choice of spouse, and the spouse affected the quality of parenting.

Studies of process could be facilitated by a greater focus on individuals and on their relationships than is customary in developmental research. It requires an attack on how individuals influence their relationships and vice versa, and on the processes (behavioral or intrapsychic) by which relationships affect each other (e.g., Dunn, 1983, pp. 804-806). This may involve substituting for the more traditional correlational approaches an- alytic methods that make more allowance for the differing impacts of a given variable on different individuals (Hinde & Dennis, 1986). And as noted above, research must cross the levels of social complexity to un- cover the dynamics of the processes involved.

RELATIONS TO OTHER APPROACHES

We have argued elsewhere that too many pates have been needlessly bloodied by investigators lighting for the honor of their paradigms, and it is not our intention to follow their example. We wish to emphasize that a relationships approach must be integrated with existing approaches, not that it should replace them.

In particular, much of what has been said here is compatible with systems theory, and with the approach of many family therapists (e.g., Minuchin, 1985). Thus our emphasis, derived in part from the orienting attitudes of ethologists (Tinbergen, 1951), on the necessity for study of successive levels of complexity, each having properties of its own not relevant to other levels, and each affected by and affecting other levels, is entirely compatible with a systems approach. We have emphasized the close interdependence of the dyadic relationships within a group or family, with influences involving complex feedback loops. We have noted the importance of mechanisms promoting and permitting change (Hinde, 1972, 1976, 1979; Hinde & Stevenson-Hinde, 1976; Stevenson-Hinde & Hinde, 1986). And a focus on relationships, and on how relationships atfect relationships, indeed involves assessing not only the individual and his/her current relationships, but also other relationships. For example, a child’s development may be affected not only by the mother/child and father/child relationships, but also by the mother/father relationship (e.g., Belsky, 1981); and past relationships may help to predict present ones with different others (e.g., Sroufe, 1985).

Yet we have hesitated to use the label of “systems theory” for a number of reasons. In the first place, it seems a little pretentious. What


we present here is hardly theory, and more generally the use of “systems theory” in this area of psychology is primarily descriptive and lacks the precision which can give it so much appeal (McFarland, 1974).

Second, in their anxiety to correct the previous bias in favor of the parts at the cost of the whole, many systems theoriests seem to swing too far the other way. It is necessary not only to conceptualize the individual “as an inter-dependent, contributing part of the systems that control his or her behavior” (Minuchin, 1985, p. 291), but also to recognize processes internal to the individual, as revealed by physiological measures (e.g., Kagan, Reznick, Clarke, Snidman, & Garcia-Coll, 1984), by ge- netic analyses (e.g., Daniels & Plomin, 1985), and by behavioral measures of individual differences. We agree that cross-situational prediction is likely to be extremely difficult (Minuchin, 1985, p. 293), but, to quote Minuchin from another context, “What is not tackled will not yield” (Minuchin, 1985, p. 294). But, as Minuchin indeed emphasizes (1985, p. 285), we see as vital an understanding of process, based on attempts to specify precisely how one relationship affects another. An attack on the nature, dynamics, and mutual influences between dyadic and polyadic relationships is a crucial step in understanding the behavior of individuals within larger systems.

CONCLUSION

To assess simultaneously the nature, dynamics, and mutual influences between relationships is a difficult but crucial step in understanding the development of an individual. The task involves the recognition of different levels of complexity, ranging from the individual to interactions, relationships, and social structure. It is necessary to be clear about what questions can be asked (e.g., about the contributions of the participants) and the nature of our explanatory concepts (e.g., temperament, attachment). Such an approach, with its roots in ethology, is compatible with a systems approach, but it hardly seems proper to reify either into the status of a theory. It is essential at this stage to remain eclectic.

REFERENCES Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patrerns of attachmeni.

A psychological study of the strange situarion. Hillsdale, NJ: Erlbaum. Alexander, R. D. (1974). The evolution of social behavior. Annual Review of Ecology and

Systematics, 5, 325-383. Andreyeva, G. M., & Gozman, L. J. (1981). Interpersonal relationships and social context.

In S. Duck & R. Gilmour (Eds.), Personal relationships (Vol. 1). Orlando/London: Academic Press.

Bates, J. E. (in press). Temperament in infancy. In J. D. Osofsky (Ed.), Handbook of infant development (2nd ed.).

Baumrind, D. (1967). Child care practices anteceding 3 patterns of preschool behavior. Ge- nefic Psychology Monographs, 75, 43-88.


Baumrind, D. (1971). Current patterns of parental authority. Developmental Psychology Monograph, 4(1, Pt. 2).

Belsky, J. (1981). Early human experience: A family perspective. Developmental Psy- chology, 17, 3-23.

Belsky, J., & Isabella, R. A. (1985). Marital and parent-child relationships in family of origin and marital change following the birth of a baby: A retrospective analysis. Child Development, 56, 342-349.

Bell, R. Q., & Harper, L. V. (1977). Child effects on adults. Hillsdale, NJ: Erlbaum. Bern, D. J., & Funder, D. C. (1978). Predicting more of the people more of the time. Psy-

chological Review, 85, 485-501. Berman, C. (1983). Differentiation of relationships among rhesus monkey infants. In R. A.

Hinde (Ed.), Primate social relationships. Oxford: Blackwell. Block, J. (1981). Some enduring and consequential structures of personality. In A. Robin, J.

Aronoff, A. M. Barclay, & R. A. Zucker (Eds.), Further explanations in personality. New York: Wiley.

Bowlby, J. (1969). Attachmen? and loss. New York: Basic Books. Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, MA: Harvard

Univ. Press. Carraher, T. N., Carraher, D. W., & Schliemann, A. D. (1985). Mathematics in the streets

and in schools. British Journal of Developmenlal Psychology, 3, 21-30. Clarke-Stewart, K. A. (1973). Interactions between mothers and their young children:

Characteristics and consequences. Monographs of the Society for Research in Child Developmenf, 38(6 & 7, Serial No. 153).

Clarke-Stewart, K. A. (1978). And daddy makes three: The father’s impact on mother and young child. Child Developmenr. 49, 466-478.

Daniels, D., & Plomin, R. (1985). Origins of individual differences in infant shyness. Devel- opmenral Psychology, 21, 118- 121.

Doise, W., & Mugny, G. (1984). Th e social developmenr of the intellect. Oxford: Pergamon. Donaldson, M. (1978). Children’s minds. London: Fontana. Duck, S., & Gilmour, R. (1981, et seq.) Personal relationships (Vols I-5). Orlando/London:

Academic Press. Dunn, J. (1983). Sibling relationships in early childhood. Child Development, 54, 787-811. Elder, G. H., Van Nguyen, T., & Caspi, A. (1985). Child Development, 56, 361-375. Endler, N. S., & Hunt, J. McV. (1968). Inventories of hostility and comparisons of the pro-

portions of variance from persons, responses, and situations for hostility and anxiousness. Journal of Personality and Social Psychology, 9, 309-315.

Endler, N. S., & Hunt, J. McV. (1969). Generalizability of contributions from sources of variance in the S-R inventories of anxiousness, Journal of Personality, 37, l-24.

Endler, N. S., & Magnusson, D. (1976). Toward an interactional psychology of personality. Psychological Bulletin, 83, 956-997.

Eron, L. D., Huesmann, L. R., Letkowitz, M. M., & Walder, L. 0. (1972). Does television violence cause aggression? American Psychologist, 27, 253-263.

Foot, H. C., Chapman, A. J., & Smith, J. R. (1980). Friendship and social relations in children. New York: Wiley.

Gelman, R., & Shatz, M. (1977). Speech adjustments in talk to 2-year-olds. In M. Lewis & L. A. Rosenblum (Eds.), Inreraction, conversation and the developmenf of language. New York: Academic Press.

Goffman, E. (1959). The presentation of self in everyday life. New York: Doubleday/An- chor.

Grossmann, K. E., Grossmann, K., Huber, F., & Warmer, U. (1981). German children’s


behaviour towards their mothers at 12 months and their fathers at 18 months in Ains- worth’s Strange Situation. International Journal of Behavioral Development, 4, 157-181.

Halverson, C. F., & Waldrop, M. F. (1970). Maternal behavior toward own and other preschool children: The problem of “owness.” Child Development, 41, 838-845.

Hanson, N. R. (1955). Causal chains. Mind, 255, 289-311. Hartup, W. W. (1983). Peer relations. In. P. H. Mussen (Ed.), Childpsychology (Vol. 4, pp.

103-197). New York: Wiley. Herdt, G. H. (1981). Guardians of the flute: Idioms of masculinity. New York: McGraw-

Hill. Hinde, R. A. (1969). Analysing the roles of the partners in a behaviour interaction-

mother-infant relations in rhesus macaques. Ann. N. Y. Acad. Sci., 159, 65 l-667. Hinde, R. A. (1972). Social behavior and its development in subhuman primates. Condon

Lectures, Eugene, OR. Hinde, R. A. (1976). Interactions, relationships and social structure. Man, 11, l-17. Hinde, R. A. (1977). On assessing the bases of partner preferences. Behaviour, LXII(lO2),

l-9. Hinde, R. A. (1979). Toward understanding relationships. Ontario/London: Academic

Press. Hinde, R. A. (1984). Why do the sexes behave differently in close relationships? Journal of

Social and Personal Relationships, 1, 471-501. Hinde, R. A., & Bateson, P. P. G. (1984). Discontinuities versus continuities in behavioural

development and the neglect of process. international Journal of Behavioral Devetop- merit, I, 129-143.

Hinde, R. A., & Dennis, A. (1986). Categorizing individuals: An alternative to linear analysis. International Journal of Behavioral Development, 9, 105- 119.

Hinde, R. A., & Herrmann, J. (1977). Frequencies, durations, derived measures, and their correlations in studying dyadic and triadic relationships. In H. R. Schaffer (Ed.), Studies in mother-infant interaction. Orlando/London: Academic Press.

Hinde, R. A. & Stevenson-Hinde, J. (1976). Towards understanding relationships: Dynamic stability. In P. P. G. Bateson & R. A. Hinde (Eds.), Growing points in ethology (pp. 451-479). Cambridge: Cambridge Univ. Press.

Hinde, R. A., Stevenson-Hinde, J., & Tamplin, A. (1985). Characteristics of 3. to 4-year- olds assessed at home and their interactions in preschool. Developmental Psychology, 21(l), 130-140.

Hinde, R. A., & Stevenson-Hinde, J., (in preparation). Varieties of sex-differences in the mother-child relationships of 42-50 month olds.

Hinde, R. A., & Tamplin, A. (1983). Relations between mother-child interaction and behaviour in preschool. British Journal of Developmental Psychology, 1, 231-257.

Humphrey, N. K. (1976). The social function of intellect. In P. P. G. Bateson & R. A. Hinde (Eds.), Growing points in ethology. London/New York: Cambridge Univ. Press.

Jolly, A. (1966). Lemur social behavior and primate intelligence. Science, 153, 501-506. Kagan, J., Reznick, J. S., Clarke, C., Snidman, N., & GarciaCoIl, C. (1984). Behavioral

inhibition to the unfamiliar. Child Development, 55, 2212-2225. Keesing, R. M. (1982). Introduction. In G. H. Herdt (Ed.), Rituals of manhood. Berkeley:

Univ. of California Press. Kelley, H. H. (1979). Personal relationships; Their structures and processes. Hillsdale, NJ:

Erlbaum. Kelley, H. H., Berscheid, E., Christensen, A., Harvey, J. H., Huston, T. L. Levinger, G.,

McClintock, E., Peplau, L. A., & Peterson, D. R. (1983). Close relationships. New York: Freeman.


Kenny, D. A. (1975). Cross-lagged panel correlation: A test for spuriousness. Psychological Bulletin, 82, 887-903.

Kenrick, D. T., & Stringfield, D. 0. (1980). Personality traits and the eye of the beholder. Psychological Review, 87, 88- 104.

Lamb, M. E. (1976). The role ofthefather in child development. New York: Wiley. Lee, C. L., & Bates, J. E. (1985). Mother-child interaction at age two years and perceived

difficult temperament. Child Development, 56, 1314- 1325. Maccoby, E. E., & Martin, J. A. (1983). Socialization in the context of the family: Parent-

child interaction. In P. H. Mussen (Ed.), Child psychology (Vol. 4, pp. I-103). New York: Wiley.

Main, M., & Weston, D. (1981). The quality of the toddler’s relationship to mother and to father: Related to conflict behavior and the readiness to establish new relationships. Child Development, 52, 932-940.

McCall, G. J. (1974). A symbolic interactionist approach to attraction. In T. L. Huston (Ed.) Foundations of Interpersonal Attraction. New York, Academic Press.

McCall, M. (1970). Boundary rules in relationships and encounters. In G. J. McCall et al., (Eds.), Social relationships. Chicago: Aldine.

McFarland, D. J. (1974). Motivational control systems analysis. Orlando/London: Aca- demic Press.

Mellen, S. L. W. (1981). The evolution of love. Oxford: Freeman. Minuchin, P. (1985). Families and individual development: Provocations from the field of

family therapy. Child Development, 56, 289-302. Mischel, W. (1973). Toward a cognitive social learning reconceptualization of personality.

Psychological Review, 80, 252-283. Osgood, C. E. (1969). On the whys and wherefores of E, P, and A. Journal of Personal and

Social Psychology, 12, 194- 199. Parke, R. D. (1979). Perspectives of father-infant interaction. J. D. Osofsky (Ed.), Hand-

book of infant development. New York: Wiley. Parke, R. D., Power, T. G., & Gottman, J. (1979). Conceptualizing and quantifying influ-

ence patterns in the family triad. In M. E. Lamb, S. J. Suomi, & G. R. Stephenson (Eds.), Social interaction anatysis. Madison: Univ. of Wisconsin Press.

Perret-Clermont, A-N., & Brossard, A. (1985). On the interdigitation of social and cognitive processes. In R. A. Hinde, A-N. Perret-Clermont. & J. Stevenson-Hinde (Eds.), Social relationships and cognitive development. Oxford: Oxford Univ. Press (Clarendon).

Pound, A. (1982). Attachment and maternal depression. In C. M. Parkes, & J. Stevenson- Hinde (Eds.) The place of attachment in human behavior. London/New York: Tavis- to&/Basic Books.

Rogosa, D. (1980). A critique of cross-lagged correlation. Psychological Bulletin, 88, 245-258.

Rubin, K. H., & Ross, H. S. (1982). Peer relationships and social skills in childhood. New York: Springer-Verlag.

Rushton, J. P., Brainerd, C. J., & Pressley, M. (1983). Behavioral development and construct validity: The principle of aggregation. Psychological Bullefin, 94, 18-38.

Rutter, M. D., Quinton, D., & Liddle, C. (1983). Parenting in two generations: Looking backwards and looking forwards. In N. Madge (Ed.), Families at risk. London: Heine- mann.

Shatz, M., & Gelman, R. (1973). The development of communication skills: modifications in the speech of young children as a function of the listener. Monog. Sot. Res. Child Devel, 38,(5).

Short, R. (1979). Sexual selection and its component parts, somatic and genital selection, as illustrated by man and the great apes. Advances in rhe Study of Behaviour, 9, 131-158.


Simpson, A. E., & Stevenson-Hinde, J. (1985). Temperamental characteristics of three- to four-year-old boys and girls and child-family interactions. Journal of Child Psychology and Psychiatry, 26,43-53.

Snow, C. (1972). Mother’s speech to children learning language. Child Developmenf, 43, 549-564.

Sroufe, L. A. (1979). The coherence of individual development. American Psychologisf, 34, 834-841.

Sroufe, L. A. (1985). Attachment classification from the perspective of infant-caregiver relationships and infant temperament. Child Development. 56, 1- 14.

Sroufe, L. A., Jacobvitz, D., Mangelsdorf, S., DeAngelo, E., & Ward, M. J. (1985). Gener- ational boundary dissolution between mothers and their preschool children: A relationship systems approach. Child Development, 56, 317-325.

Stevenson-Hinde, J., & Hinde, R. A. (1986). Changes in associations between characteristics and interactions. In R. Plomin and J. Dunn (Eds.), The srudy of temperament: changes, continuities and challenges. Hillsdale, NJ: Erlbaum.

Stevenson-Hinde, J., (in press). Towards a more open construct. In D. Kohnstamm (Ed.), Temperament discussed. Holland: Swets & Zeitiinger.

Sullivan, H. S. (1938). The data of psychiatry. Psychiatry, 1, 121-134. Tinbergen, N. (1951). The study ofinstinct. Oxford: Oxford Univ. Press (Clarendon). Whiting, B. B., & Whiting, J. W. M. (1975). Children ofsix culrures. Cambridge, MA: Har-

vard Univ. Press. Williams, J. E., & Best, D. L. (1982). Measuring sex stereotypes. Beverly Hills, CA: Sage. Youniss, J. (1980). Parents and peers in social development. Chicago: Univ. of Chicago

Press.

RECEIVED: October 16, 1985; REVISED: January 10, 1986

interpersonal relationships and child development

Documents